Thursday, July 16, 2009

The Value of Bobby Abreu

According to Fangraphs, he's already earned more than double his 5 million 2009 salary, at 10.5 million. He hasn't hit too many homers, but with a .396 OBP, 19 steals, and 58 RBI, he's been a joy to watch. His value goes beyond the stats though, as Torii Hunter gives him some credit:

Having Bobby Abreu here has been big for me. I'm more disciplined at the plate than I've ever been, and I can thank Bobby for that. He's a master up there, and he's a great guy to play with, because he's so willing to share his knowledge. He's also a really funny guy, helping keep things loose.


With 33 walks so far, Torii has the best walk rate of his career, and getting better pitches through improved patience sure hasn't hurt his power or batting average.

Saturday, July 11, 2009

Top 500

The top 300 lists have been replaced by a top 500 list.

Timelining

Are the players of today better than players of the past? If so, how much better? These are not easy questions to answer. It is possible to construct a study to estimate it. David Gassko did such for The Hardball Times two years ago, and did a good job, producing very reaonable results. Unfortunately, it is far from conclusive and very sensitive to whatever assumptions you use to set it up.

The approach, which I have played around with, is to compare how players do from one season to the next. But how much of it is age decline, and how much is the change in league strength? You really can't tell, since the same process we're using here is what we use to compute aging factors in the first place.

It is necessary to regress to the mean as well. Take two average players, but one is +15 by luck and the other is -15 by bad luck. The next year, player 1 should be average. But player 2 may not play, or else be limited to a bench role, so as a group you would see decline even if there was no real change at all.

Some criticized David's study for regressing too much, including some Baseball Prospectus Authors, who apparently didn't see the need for regression at all. They are demonstrably wrong about not needing regression*, but may be right in that David did too much. It's pretty much impossible to tell.

I tried to get around the regression issue by choosing only players who were average in year one, or very close to it. Even with that, what age ranges should I include? I tried using ages 26-28 in year one (27-29 year 2). This would work if it were true that on average, 28 year olds were equal to 27 year olds. I also tried looking at ages 25-28 and 26-29, which would look at, on average, ages 26.5 and 27.5. This works if age 27 is indeed the true peak. The difference in results is tiny, but become huge when chained to look at seasons 100 years apart.

Another issue that isn't even touched is improvements that affect all players. We're assuming that the player is constant from year 1 to year 2 and any changes measured represent the league changing around him. What if some new, illegal or legal, nutritional supplement makes every hitter and pitcher better? This study would not pick that up at all. I'm not sure it even matters. If there are great advancements making everyone a better player then we should not penalize great players from the past for not having these advantages. If they played today, they would have them just like everyone else. I think we want to look at the greatness of the player, not what improved circumstances allow him.

One thing that can should be considered is game improvements from using a greater pool of the best available players. Major Leaguers before 1947 would not have dominated their leagues to the same extent if they had not excluded players of darker skin color.

*Honus Wagner is the case in point. The no-regression time lines would make him a replacement level hitter (and Ty Cobb not much better I guess) in today's game. Here are some factors that determine how great a ballplayer is, in order from the ones that seem to have changed the most over the years to the least:

1. Size and Strength
2. Speed
3. Throwing Ability
4. Hand/eye coordination, reactions.

No question that players of today are much bigger and stronger than players 100 years ago. But Wagner was an exception, 5'11, 200 pounds, and was one of the earliest players to work out with weights. He would not be out of place in today's game. I'm not sure how his speed would rate today. His arm would be one of the best among shortstops, as Wagner is known to have made some incredible long distance throws, I believe near 400 feet. I don't think it is possible, even under the most favorable circumstances, to throw a baseball that far without being able to throw at least 90 MPH. I can't think of any reason why #4 would change that much, unlike muscle mass and speed throw better nutrition. Wagner was the best of his day in this regard, and would almost certainly at least be among the best today.

Monday, July 06, 2009

Best Player, pound for pound

This post got me thinking. Is Ichiro the best player pound for pound?

The answer is no, it's still Albert Pujols. But Ichiro does come in second. I took each player's career wins over replacement, and converted to a rate stat, wins per 12000 plate appearances. In other words, the number of PA a great player would put up in a long career.

Albert is the best player, of course, at 152 WAR per 12000. Ichiro does rank very high among active players, about 16th place at 93 WAR. Divide this by weight and we get the best pound for pound players:


WAR/Wt Player
0.659 Pujols
0.578 Ichiro
0.565 A-Rod
0.558 Mauer
0.529 H Ramirez
0.522 Utley
0.502 Sizemore

Sunday, July 05, 2009

Updated Wins Above Replacement

I've made a few changes:

1. Add a column for reaching on errors for batters.

2. Cap the replacement column, which is based on plate appearances, at 4 PA per game. The batter will be evaluated as if he has the lessor of 4 PA per game, or his actual plate appearances. This eliminates the leadoff bonus, where leadoff hitters may have added 1-2 runs per year to their rating just because they bat at the top of the lineup and get more bats. One reason for this is that in evaluating runs over replacement, you have to assume that the replacement player will not bat at the top of the order, but the bottom, with the other hitters moving up.

3. Pitcher's hitting and pitching records are all on the same page.

4. Pitchers have a new set of columns, showing how far above/below average they were in several independent categories. This is meant not as a value measure, but more a descriptor. Not "How great a pitcher was he?", but "What kind of pitcher was he?" This shows that practically all of Randy Johnson's value came from his strikeouts, he was essentially an average pitcher if he didn't whiff you. Roger Clemen's value lies more in a mix of strikeouts and homerun prevention. Tom Glavine, on the other hand, was below average in strikeouts but excelled in keeping the ball in the park and stranding runners.

5. Pitcher's hitting records include a position adjustment, so his WAR Total shows how valuable he was relative to the average pitcher. (For pitcher's hitting, average and replacement level are the exact same thing.)

6. And finally, the numbers go all the way back to 1871 for hitters and 1876 for pitchers. Some of the estimates used to fill these stats in, like the baserunning regression formula or the JAARF fielding numbers, are not to be trusted as anything more than a reasonable guess. Catcher defensive ratings are based on passed balls and errors only. It is not worth it to even try to estimate performance against the running game by catcher assists. Mike Piazza had about as many assists per game as Johnny Bench. Enough said.

7. The 300 lists have not been updated. When I do, it will probably be a 500 list since so many more players are added.

Friday, July 03, 2009

Highest Leverage Index of All Time

...Or at least 1954, the years retrosheet has play by play files for.

I sorted my career pitcher log by career leverage index, which measures the volatility of a game (1 is average, 9th inning, 1 run lead is high, 7th inning, 12 run lead is low). Since there are some pitchers who came up for a cup of coffee, might have for some emergency found themselves in a crucial sitation, and never pitched again, I was expected to see a lot of 2-5 inning guys before I got to the real careers.

In reality, there are only a few of them. Bruce Sutter ranks 8th with a 2.0 leverage index, behind 7 guys who pitched no more than 3 innings each. A few pitchers come in at 1.9, K-Rod, Percival, John Franco, and Trevor Hoffman among them. Mariano Rivera is at 1.8.

Only one pitcher has a leverage index of 3.0, hitting that on the nose. Surprise is, it wasn't even a real pitcher, but catcher Brent Mayne for one inning on August 22, 2000. The game between the Rockies went 12 innings, and the Rockies had already used 9 pitchers before handing the ball to Mayne. I don't remember the whole story, maybe the last guy got hurt and Mayne had to pitch since there was nobody left. The 6 pitchers before him didn't work very hard, throwing between 3 and 12 pitches each.

Anyway, Mayne faced Tom Glavine and got a ground out. Walt Weiss flew out to center. This pitching stuff ain't that hard, is it? A single, wild pitch, and walk later, Mayne found himself staring down the reigning NL MVP, Larry Wayne Jones, who had already homered in the game. No big deal, Jones grounded out to third, the Rockies won the game in the bottom of the inning, and Mayne's legend as the most leverage pitcher of all time was established.